We propose an intelligent content based image retrieval system and it is an extension of a previously published research paper (S. Kulkarni et al., 1999), where a neuro-fuzzy technique was presented. The CBIR system will accept multiple queries as input such as "mostly red and many blue and few green" that can be provided online and the outputs of the system are the images with their confidential values. The system uses fuzzy logic to interpret multiple natural expressions such as mostly, many and few and a neural network to learn the meaning of mostly red, many red and few red. The system has been implemented on the World Wide Web using CGI scripts and C programming language. Previously (S. Kulkarni et al., 1999), we presented some preliminary results using a single query. The paper presents the extension of our online system for multiple queries such as "many blue and few green". A number of experiments using our online system is conducted and the results are included and discussed in the paper. © 1999 IEEE.